-
Notifications
You must be signed in to change notification settings - Fork 1
/
viterbi.py
142 lines (112 loc) · 3.82 KB
/
viterbi.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
from __future__ import division
from __future__ import print_function
from collections import Counter
from nltk.corpus import brown
from nltk.util import ngrams
from nltk.tokenize import sent_tokenize, word_tokenize
from nltk.tokenize import RegexpTokenizer
from nltk.data import load
import nltk
tagdict = load('help/tagsets/brown_tagset.pickle')
taglist = tagdict.keys()
#taglist = ['NOUN','ADJ','ADP','ADV','CONJ','DET','NUM','PRT','PRON','VERB','.','X']
taglist_size = taglist.__len__()
tag_sequence_corpus = brown.tagged_sents(tagset='brown')
tag_list = []
corpus_with_tag = []
#Everything is tagged, including the punctuations and the lines
print("Creating tag lists....")
for sentences in tag_sequence_corpus:
for tags in sentences:
tag_list.append(tags[1])
corpus_with_tag.append((tags[0],tags[1]))
print("Done creating tag lists....")
print("Creating tag corpus...")
#Code snippet that works upon the unigrams list
unigrams = ngrams(tag_list,1)
unigrams_freq = Counter(unigrams);
#Code snippet that works upon the bigrams list
bigrams = ngrams(tag_list,2)
bigrams_freq = Counter(bigrams);
#Code snippet that works upon the trigrams list
trigrams = ngrams(tag_list,3)
trigrams_freq = Counter(trigrams);
#Length of the corpus
len_corpus = brown.words().__len__()
word_with_tag = Counter(corpus_with_tag)
print("Corpus tagged!")
def S(k):
"This function returns the set value S for the viterbi algorithm"
if k in (-1,0): return [""]
else: return taglist
def argmax(ls):
return max(ls, key = lambda x:x[1])
def trigramCounter(w,u,v):
ans = 0.0 if bigrams_freq[(u,v,)] == 0 else float(trigrams_freq[(w,u,v)])/float(bigrams_freq[(u,v,)])
ans += 0.0 if unigrams_freq[(u,)] == 0 else float(bigrams_freq[(u,v,)])/float(unigrams_freq[(u,)])
ans += float(unigrams_freq[(v,)])/float(len_corpus)
ans = ans/3;
return ans
def q(v,w,u):
"This function returns the trigram count estimation"
return trigramCounter(w,u,v)
def e(x,u):
w_t = word_with_tag[(x,u,)]
t = unigrams_freq[(u,)]
if t == 0 : t = 1
return w_t/t
def Viterbi(sentence):
"This function implements the viterbi algorithm for a given function"
#The pi refers to the dictionary for the Viterbi tagset probabilties
pi = {}
#Initialization
pi[0,"",""] = 1.0
#This array, called as backpointer is used to retrieve the tags corresponding to a given sentence
bp = {}
#Tokens converts the sentence into array of words and punctuations
tokens = word_tokenize(sentence)
n = tokens.__len__()
#Padding so that the sentence begins at the position 1
tokens = [""]+tokens
#The viterbi algorithm
for k in range(1,n+1):
for u in S(k-1):
for v in S(k):
bp[k,u,v], pi[k,u,v] = argmax([(w, pi[k-1,w,u]* q(v,w,u) * e(tokens[k],v)) for w in S(k-2)])
#Now the dictionary pi consists of the maximum probabilities of tag sequences
#We first create an array of n+1 length to store all the tags
y = [""]*(n+1)
(y[n-1],y[n]),score = argmax([( (u,v), pi[n,u,v]*q(".",u,v) ) for u in S(n-1) for v in S(n)])
for k in range(n-2,0,-1):
y[k] = bp[k+2, y[k+1],y[k+2]]
y[0]=""
return y
def cutit(s,rem,n):
n = n + rem.__len__()
return s[n:]
def main():
while 1 == 1 :
print("Enter a statement")
statement = raw_input()
statement = statement.lower()
f = open('eliminate.txt','r')
removable = []
for line in f:
removable.append(line[:-1])
for k in range(0,removable.__len__()-1):
if(removable[k] in statement):
statement = cutit(statement,removable[k],statement.find(removable[k]))
tagged_arr = Viterbi(statement)
#check if all of the elements are same
count = 1
tag = tagged_arr[1]
for i in range(2,tagged_arr.__len__()):
if tagged_arr[i] == tag:
count = count + 1
if count == tagged_arr.__len__()-1:
tokens = word_tokenize(statement)
n = tokens.__len__()
for i in range(0,n):
tagged_arr[i+1] = Viterbi(tokens[i])[1]
print(tagged_arr)
main()